One of the most pervasive problems for stroke survivors is movement deficits. Recent research strongly supports prolonged practice of functionally-relevant activities of the upper limb, even though therapy time is quite limited by the current medical economic system. This grant focuses on new developments in human- robot interactions (haptics) that have revealed prospects in the areas of motor teaching and rehabilitation. Specialized robotic devices combined with computer-displays can tirelessly exert force, augment feedback, and redirect error in order to speed up, enhance, or trigger the motor relearning process. These approaches could extend and greatly enhance the recovery process. The first strategy that often comes to mind for teaching movements is to guide the limb along the desired path. However, a promising alternative approach is to make movements more difficult by deflecting them from the desired path. People develop, through practice, the ability to counteract forces that distort the mechanical world, and if these forces are properly designed and applied, a desired movement pattern occurs when the forces are eventually switched off. We and others have also obtained similar results by distorting the visual world using prisms or virtual reality displays. In these studies, the subject sees something unexpected that is perceived as an error. Our results point to a single unifying theory: Errors induce learning, and judicious error augmentation (through forces or visual distortions) can lead to lasting desired changes. Interestingly, this process appears to bypass conventional learning mechanisms that require intense concentration - results are the same if the subjects have a conversation or listen to music. They often consider it a game. Until now very little of this research has been functionally relevant because the devices'ranges of motion were small, were two dimensional, and were lacking an appropriate visual interface. Three dimensional movements introduce the daunting new challenge of gravitational effects that could reduce (or perhaps heighten) the potential of error augmentation training. Our lab has spent several years developing a large- workspace, three dimensional haptics/graphics system. The aims of this grant are to build on our promising body of evidence and expand our error augmentation training work to a large workspace in three dimensions. Accordingly, the experiments below further refine our understanding of error augmentation(Aim 1), expand our approaches to three dimensions (Aim 2), and then move towards clinical application by testing our approaches on stroke survivors (Aim 3).